OUP user menu

The dreamy state: hallucinations of autobiographic memory evoked by temporal lobe stimulations and seizures

Jean-Pierre Vignal, Louis Maillard, Aileen McGonigal, Patrick Chauvel
DOI: http://dx.doi.org/10.1093/brain/awl329 88-99 First published online: 30 November 2006


Using results from cortical stimulations, as well as the symptoms of spontaneous epileptic seizures recorded by stereoelectroencephalography we re-studied the phenomenon of the dreamy state, as described by Jackson (Jackson JH. Selected writings of John Hughlins Jackson. Vol 1. On epilepsy and epileptiform convulsions. Taylor J, editor. London: Hodder and Stoughton; 1931). A total of 15 sensations of déjà vécu, 35 visual hallucinations consisting of the image of a scene and 5 ‘feelings of strangeness’ occurred. These were recorded during 40 stimulations in 16 subjects, and 15 seizures in 5 subjects. Forty-five per cent of dreamy states were evoked by stimulation of the amygdala, 37.5% by the hippocampus and 17.5% by the para-hippocampal gyrus. During both spontaneous and provoked dreamy state, the electrical discharge was localized within mesial temporal lobe structures, without involvement of the temporal neocortex. Early spread of the discharge to the temporal neocortex appeared to prevent the occurrence of the dreamy state. Semiological analysis showed a clinical continuity between déjà vécu and visual hallucinations, the latter often consisting of a personal memory that was ‘relived’ by the subject; such memories could be recent, distant or from childhood. With one exception, the particular memory evoked differed from one seizure to another, but were always drawn from the same period of the subject's life. Given the role of the amygdala and hippocampus in autobiographic memory, their pathological activation during seizures may trigger memory recall. This study of the dreamy state is in keeping with other evidence demonstrating the constant and central role of the amygdala and hippocampus (right as much as left) in the recall of recent and distant memories. It demonstrates the existence of large neural networks that produce recall of memories via activation of the hippocampus, amygdala and rhinal cortex.

  • dreamy state
  • autobiographic memory
  • amygdala
  • hippocampus
  • cortical stimulation


Jackson described elaborate psychic states in epileptic subjects, which he grouped as ‘dreamy state’, and which were distinguishable from elementary epileptic sensations. The ‘dreamy state’ refers to a sensation of déjà vécu and/or complex visual hallucinations (e.g. of scenes, faces or people) and sensation of ‘strangeness’. Jackson localized this in the mesial temporal lobe (MTL).

Penfield (1967) distinguished between feelings of familiarity (déjà vécu) and elaborate visual hallucinations. He classed feelings of familiarity and strangeness as illusions, grouping these with other sensory and ‘emotional’ illusions. The denomination ‘interpretive illusions’, though not currently considered valid, was used to describe these (Penfield and Perot, 1963). Hallucinations were considered the replaying of past experience, termed ‘experiential hallucinations’, and presumed to be based only upon personal memories. Production of feelings of familiarity and visual hallucinations was elicited by electrical stimulation sites distributed widely over the temporal neocortex.

The dreamy state could be provoked by stimulation of the MTL structures (Mahl et al., 1964; Horowitz et al., 1968; Ferguson et al., 1969). Grouping together the mental phenomena provoked by hippocampal and amygdalar stimulations, Halgren et al. (1978) noted that the presence or absence of widespread afterdischarge appeared more important than stimulation site. Dreamy state was produced only with recruitment of several brain structures, and its content reflected the subject's personality.

Gloor et al. (1982) found that experiential hallucinations could be evoked by limbic stimulation either without after-discharge or with after-discharge restricted to MTL structures. Complex visual hallucinations were only provoked by stimulation of MTL structures, preferentially the amygdala. These responses reflected both the functional role of the region stimulated and the individual past experience of the subject. Bancaud et al. (1994) analysed the dreamy state according to Jackson's definition and emphasized that the afterdischarge involved both MTL structures and temporal neocortex; these authors directly linked the evoked hallucinations to mechanisms of declarative memory.

All of these previous authors considered these manifestations too elaborate to be explained by involvement of a single cerebral structure. For Penfield (1967), experiential hallucinations reflected a bi-directional activation of temporal neocortex (memory storage) and centrencephalic system (memory integration). Respective contributions of the MTL structures and temporal neocortex remain the subject of debate. A recent stimulation study (Bartolomei et al., 2004) emphasized the role of the entorhinal cortex in déjà-vu, and the peri-rhinal cortex in the reminiscence of memories.

The definition of the term ‘experiential’ itself is the subject of debate (Gloor, 1990; Wolf et al., 2000). These phenomena are heterogeneous, encompassing complex auditory hallucinations and visual illusions (Blume et al., 1993).

This work aims to delineate the definition and the significance of the dreamy state, to determine the localization of the ictal discharge and to discuss its relationships with the mechanisms of memory.

Material and methods

In the current study, the ‘dreamy state’ refers to the definition of Jackson including déjà-vu, reminiscence of memories and sensation of ‘strangeness’. We have not included vaguely described or poorly defined sensations.

Among a population of 180 subjects undergoing stereoelectroencephalography (SEEG) between 1990 and 1997 in the Epilepsy Unit of Rennes University Hospital, 17 patients described 55 dreamy states (Table 1). The 17 SEEG recordings were carried out using a total of 133 electrodes (83 temporal and 50 extratemporal electrodes) (Table 2).

View this table:
Table 1

Patient characteristics

Patient no.AgeDominant handWada*MRIEpileptogenic zoneMedical historyDreamy state during seizures
141RightGanglioma of right MTL and temporal poleRight MTLDéjà vécu
217RightLeft TFNo
349RightLeft hippocampal sclerosisLeft and right TLNo
427RightRight MTLDéjà vécu
542RightRight hippocampal sclerosisRight MTLDéjà vécu
642RightLeftLeft hippocampal sclerosisLeft MTLFebrile convulsionsv.h.
731RightRight MTLv.h.
828LeftLeftRight prefrontalno
922LeftLeft TFno
1023RightLeft and right TLFebrile convulsionsv.h.
1133RightDNET of right superior temporal gyrusRight frontalno
1218RightDNET of left fusiform and para-hippocampal gyrusLeft temporo-occipitalBorn prematurelyDéjà vécu and v.h.
1324RightLeft hippocampal sclerosisLeft and right TLFebrile convulsionsFeeling of strangeness
1418LeftLeftLeft cingulate gyrus tumourLeft prefrontalNo
1535RightTemporal pole DNETRight temporal (mesial and lateral)v.h.
1627LeftLeftLeft temporo-occipitalDream?
1728LeftLeftRight fusiform gyrus DNETRight MTLv.h.
  • DNET = dysembryoplastic-neuroepithelial tumours; v.h = visual hallucination; MTL = mesial temporal lobe; TL = temporal lobe; temporo-frontal.

  • *Dominant hemisphere for speech by the Wada test.

View this table:
Table 2

Main temporal lobe structures explored by one or more electrodes (number of patients)

  • Am = amygdala; H = hippocampus; Phg = parahippocampal gyrus; Stg = superior temporal gyrus.

The SEEG method involves the stereotactic placement of intracerebral electrodes consisting of 10–15 contiguous contacts (Talairach et al., 1974). The goal of the SEEG is to delineate the ‘epileptogenic zone’ (Chauvel et al., 1996). The electrode implantation sites were chosen based upon non-invasive data from an earlier phase of investigation, including clinical semiology, surface video-EEG and MRI. The electrodes, which were orthogonally implanted within the temporal lobes, simultaneously explored mesial and neocortical temporal structures (Fig. 1).

Fig. 1

Stereo-EEG schema of Patient 12 that shows a left temporo-occipital depth exploration.

SEEG recording lasted from 5 to 10 days, depending upon whether or not ictal recordings had been obtained in the first 5 days. Patients were interviewed and examined during the course of their seizure by a doctor or nurse trained in epileptology. Electrical stimulation of the cerebral cortex took place from the third day of recording onwards. Stimulation sessions lasted a maximum of 2 h. The results of these stimulations contribute to the localization of the epileptogenic zone (Chauvel et al., 1993). Patients were interviewed and examined both during and after stimulations. Bipolar stimulations were carried out between two adjacent electrode contacts. The stimulus is a biphasic rectangular wave pulse, with each pulse lasting 1 ms. These are applied in a 5 s train at a frequency of 50 Hz. Two different stimulators were used in the course of this study: in the first, the amplitude could be varied between 2.5 and 6 V (used for Patients 1–5) and in the second, the current strength could be varied from 1.5 to 3 mA (used in Patients 6–15). For Patient 10, two dreamy states were provoked by a stimulation carried out at 1 Hz.

All seizures were recorded and stored using S-VHS band video, and the EEG data archived both in paper and digital formats. All data were analysed at the time of the original exploration and then reviewed for the purpose of this study by one of the authors.


A total of 40 dreamy states were provoked by stimulation, in 16 subjects. Fifteen dreamy states occurred spontaneously, in 5 patients. In 4 patients, both spontaneous and provoked dreamy states were recorded.

Dreamy states provoked by stimulation (Table 3)

Description and classification of the dreamy state

A sensation of déjà vécu was reported after 14 stimulations in 6 patients. Patients used both the terms ‘déjà vécu’ and ‘déjà vu’ to describe this same phenomenon. Four out of 6 patients described a dreamy state during spontaneous seizures, of which 3 were déjà vécu and one was a feeling of strangeness.

View this table:
Table 3

Dreamy states provoked by stimulation

Number (patient number. seizure number)Intensity of stimulation V = volts mA= milliamperesAnatomical structure stimulatedResponseDuration of afterdischarge (s)Location of afterdischargeSpread of afterdischarge
1.13.5 VR hDéjà vécua80am, h, phgNo
1.23 VR amDéjà vécub30am, h, phgMtg, stgc
2.36 VR amDéjà vécu60amno
3.44 VR hScene: reliving a parachute jump40Am, h, phgno
4.52.5 VR amDéjà vécu40Am, h, phgNo
4.63 VR amDéjà vécuNoAm, h, phgNo
4.73 VR hDéjà vécu35Am, h, phgNo
4.84 VR phgDéjà vécu15Am, h, phgNo
4.94 VR hDéjà vécu40Am, h, phgNo
4.104 VR amDéjà vécu40Am, h, phgNo
5.114 VR amFamiliar television advertisementNo
5.123 VR hImpression of being elsewhereNo
5.134 VR phgDéjà vécuNo
5.144 VR amDéjà vécu10Am, h, phgNo
6.152 mAL amCountryside scene from childhoodd70Am, phgNo
6.162 mAL phgCountryside scene from childhood90Am, phgNo
7.172 mAR amScene from childhoode40AmNo
7.183 mAR amScene from childhoodf60Am, phgNo
8.191.5 mAR amScene and déjà vécu10Am, acgNo
9.202 mAL amImpression of being someone else30Am?No
9.212.5 mAL amImpression of leaving his body30Am?No
10.221.2 mAL hNight-time sceneg220Am, h, phgNo
10.234 mAR hSceneh160Am, h, phg, pcgL am, h, mtg and stgi
11.242 mAR amDéjà vécu60Am, hNo
11.252.6 mAR amVision of a television showj32Am, hNo
12.261.5 mAL hImpression of being elsewhere100Am, h, phgNo
12.271.5 mAL amScene: dark person speaking to himNo
12.281.5 mAL amFamiliar dream10Am, h, phgNo
12.291.5 mAL phgScenek20Am, h, phgNo
13.301 mAR hDéjà vécu8hNo
13.311.5 mAR hDéjà vécu8hNo
14.321.5 mAL amFamiliar character from a film80Am, h, phg, acgNo
14.331.5 mAL hIll-defined, unpleasant vision40Am, h, phgNo
14.341.5 mAL phgFamiliar vision that he is unable to describe20Am, h, phg, acgNo
14.351.5 mAL phgVision of a familiar face40Am, h, phgNo
14.361.5 mAL phgVision of a familiar person40Am, h, phgNo
15.371.5 mAR hRecent scenel80Am, h, phgNo
15.381.5 mAR hRecent scene that he is unable to describe
16.391.5 mAL hRelives the scene, though not the normal experience50Am, h, phgNo
16.401.5 mAL hDream4hNo
  • R = right; L = left; am = amygdala; h = hippocampus; phg = para-hippocampal gyrus; mtg = middle temporal gyrus; stg = superior temporal gyrus; acg = anterior cingulate gyrus; pcg = posterior cingulate gyrus; ifg = inferior frontal gyrus.

  • a‘It's like in my seizures, I'm reliving something … but I can see you clearly … . It's as if what is happening now has already happened to me, it's like an old memory that I am in the middle of living out’.

  • b‘It's like yesterday evening … I have the impression that everything around me has been here before, that it has already happened, I feel as if I am going backwards in time’.

  • cDischarge spread 15 s after the start of the discharge.

  • d‘It's exactly the same as the last time, it's all come back to me … it's behind, it's always thoughts from childhood, it's always visual, it's a place behind the house, the field where my father put his car, near a lake … It's not always the same countryside; I've forgotten the story of this countryside … Yes, it's pleasant because we were going to get the car from behind the house, it's a happy memory, it's never unpleasant.’

  • eVision of a bald man dressed in black, coming towards her from behind; associated with a feeling of imminent death; she is pale, with piloerection. She is re-experiencing an experience of anaesthesia by facemask during a tonsillectomy at the age of 14 years.

  • fSame scene as e.

  • g‘I'm starting to see lots of things, loads of people … it's still vague and strange. I've got an initial picture, a memory … I feel locked in again, something to do with the evening, the night … it's strange … . like after a party, sad things … there is a mixture of last night and this morning … ’ These remarks were peri-ictal. After the end of the discharge, there was complete amnesia.

  • h‘It's starting, it's rising up to my eyes … I'm always frightened of something … I feel something, like in dreams, I always see people, loads of people, people that I see in the hospital … ’.

  • iFifty eight seconds after the onset of the discharge in the limbic structures there is spread into the amygdala and left hippocampus followed by the temporal neocortex.

  • jWell-known game from a popular French television station (‘Millionaire’).

  • kDistant, unpleasant memory from sitting an exam that was supervised by a woman.

  • l‘I see myself playing the drums, with people from my family listening to me’.

In 11 patients, 21 stimulations provoked a mental hallucination of memory with a prominent visual component. The visual nature of the phenomenon was routinely confirmed. These included memories from childhood, and memories of a television advert or programme. The mnesic nature of the phenomenon was its most frequent feature and clearly confirmed by the patients themselves. More rarely, the visual hallucination did not have an obviously mnesic character, but the association between visual hallucination and the sensation of déjà vécu tended to give patients a feeling of reminiscence. When these memories were sufficiently clear, they were placed within a time context, being either distant or recent memories.

Usually, the patient's state of ‘mental diplopia’ during the hallucination was apparent from their remarks: subjects tended to describe the visualized scenes as if they were spectators. Certain patients (Patients 10, 12 and 14) experienced their hallucinations particularly when the dreamy state occurred concomitantly with a prolonged afterdischarge, the hallucination and the sensation of déjà vécu being intricately associated. For this reason, the description of symptoms could be rather ambiguous. The content of the hallucination varied within the same subject during the course of stimulation. Only Patient 7 always reported the same hallucination both after stimulation and during spontaneous seizures. This hallucination was a precise recollection of a traumatic scene from his childhood. Seven out of 11 patients reported a dreamy state during their seizures. In 6 of these cases, the descriptions of these dreamy states were identical to those obtained with stimulations. One patient described déjà vécu during his seizures (Patient 5).

Five stimulations provoked a phenomenon that we have termed ‘feeling of strangeness’, in 4 subjects. Three of these patients described the somewhat different phenomenon of dreamy state during their spontaneous seizures, either déjà vécu or mental hallucination.

Stimulation sites

Forty-five per cent of dreamy states were provoked by stimulation of the amygdala, 37.5% by stimulation of the hippocampus and 17.5% by stimulation of the para-hippocampal gyrus.

When stimulating the MTL, 22 right-sided stimulations in 14 patients, and 16 left-sided stimulations in 8 patients provoked a dreamy state. For Patient 10, visual hallucination was provoked by stimulation of, first, the left hippocampus and, then, the right hippocampus. None of the left-sided stimulations gave rise to a déjà vécu sensation.

Anatomo-electroclinical correlations during stimulations

Thirteen out of 14 sensations of déjà vécu occurred simultaneously with an afterdischarge localized to MTL. Stimulation of the para-hippocampal gyrus did not provoke any after-discharge. After 9 stimulations in 3 patients, the after-discharge extended throughout the amygdala, hippocampus and para-hippocampal gyrus. The amygdala alone was involved in the after-discharge in one patient, and the amygdala and hippocampus together in another patient. In one subject, the hippocampus was the only structure of the MTL explored by SEEG (Patient 13). In 12 out of 13 stimulations there was no spread into the temporal neocortex; neither the middle nor the superior temporal gyri were involved in 5 of the 6 patients who had electrodes placed within these structures.

Eighteen out of 21 visual hallucinations provoked by stimulation occurred simultaneously with a discharge in the MTL. On one occasion, a spread into the contralateral MTL was seen (Patient 10). In one case there was a late propagation (58 s following stimulation) into the temporal neocortex. Three stimulations were not accompanied by any discharge: 2 of these were stimulations of the amygdala and 1 of the hippocampus.

The ‘feelings of strangeness’ were accompanied by a discharge on 4 occasions out of 5. Two after-discharges were localized to the amygdala and 2 involved the amygdala, hippocampus and para-hippocampal gyrus.

Associated signs and state of consciousness (Table 4)

A subjective feeling of fear was the symptom most often associated with dreamy state.

View this table:
Table 4

Symptoms and signs associated with the dreamy state evoked by stimulations and spontaneous seizures

None17 (9)2 (2)
Fear, anxiety13 (6)6 (1)
Pallor7 (4)8 (2)
Hot flush4 (2)
Rubefaction4 (2)4 (2)
Epigastric sensation3 (3)4 (2)
Nausea, vomiting3 (3)
Cold, piloerection2 (1)5 (1)
Diffuse pins and needles2 (1)
Thoracic sensation2 (2)
Secondary loss of contact2 (2)1 (1)
Chewing2 (2)1 (1)
Gestural automatisms2 (2)1 (1)
Emotional manifestations7 (3)
  • Number of patients showing that sign in parentheses.

Two stimulations in two different patients provoked chewing movements associated with automatisms. In both of these cases a loss of contact occurred and there was a widespread discharge throughout the lateral temporal cortex.

Seven stimulations in three patients provoked an emotional modification other than fear: these included joy, sadness, aggressive behaviour and over-familiarity with the examiner.

It was difficult to ascertain the state of consciousness during the dreamy state. Attempts were made to assess this during conversation between the patient and physician. Our examination focused on determining the content of the dreamy state. Mnesic capacity was assessed based on the patient's ability to reproduce the content of hallucinations. We were able to reliably obtain descriptions of patients' ictal experiences in 5 cases. Two patients were capable of completely reproducing the content of their dreamy state, in one case this being always exactly the same hallucination. In cases where the hallucination was very brief, memory for this did not appear to be altered. Three patients experienced a prolonged hallucination; the description obtained immediately at the end of these tended to be rather sparse. After a delay following the hallucination (2 h later, or the next day), the content of the hallucination tended to have been forgotten or was reported in a vague manner.

Correlations between the epileptogenic zone, stimulation site and the discharge

In 12 patients, the epileptogenic zone included the mesial temporal region where the stimulation sites and after-discharge were confined. In 3 patients who had an epileptogenic zone localized to the prefrontal cortex, stimulation of MTL gave rise to a dreamy state. For Patient 2, stimulation was performed in the right amygdala, contralateral to the epileptogenic zone in the left fronto-temporal region.

Spontaneous dreamy states

Fifteen spontaneous dreamy states were recorded in 5 patients (Table 5).

View this table:
Table 5

Spontaneous dreamy states

Patient identification numberNumber of seizures with dreamy stateDescriptionDuration (s)Localization of discharge*Hemisphere of onsetSpread of discharge**
51Déjà vécu70Am, h, phgRightYesa
64Pleasant scenes from childhoodb30–75Am, phgLeftNo
76Frightening scenes from childhoodc90–120Am, phgRightYesd
171Recent pleasant scenesfH, phgRightNo
  • *During dreamy state; **involving temporal and frontal neocortex.

  • aTowards middle temporal gyrus and prefrontal cortex 15 s after the start of the discharge.

  • bThese were scenes from childhood that varied in content from one seizure to another. Seizure 6.1: ‘It's visual. I can see you, I am completely aware; I know exactly what I'm saying even as I know inside that I see something. Memories, always from childhood, always outdoor scenes, the countryside … I was reading a book that was saying something about children and I saw the little farmhouse where I went to look for frogs when I was little.’

  • cAlways the same scene: anaesthesia by mask at the age of 14 years during a tonsillectomy.

  • dDuring 3 out of 6 seizures, >60 s after the onset of the discharge in the amygdala.

  • eScenes from long ago accompanied by a sensation of déjà vécu. These are described in a confused manner; there are individuals present, one of whom is in a gymnasium.

  • f‘A landscape, with loads of people, a moving crowd; it's always a recent scene, never the same.’

Description of the dreamy states and associated symptoms

For 4 patients, 14 dreamy states were described as visual hallucinations, the content of which evoked memories that could be recent, remote or from childhood. In one patient, the phenomenon was limited to a sensation of déjà vécu.

The same dreamy states were also provoked by stimulation on 8 occasions in 4 patients: 5 stimulations of the amygdala and 3 stimulations of the parahippocampal gyrus. On 3 occasions, in 2 patients, a different dreamy state from their normal spontaneous symptom was provoked by stimulation (Patients 5 and 12).

The associated symptoms (Table 4) were principally those of fear, epigastric sensation and autonomic signs.

Anatomo-electroclinical correlations: spontaneous dreamy states

The onset of the ictal discharge was always within the MTL; on 12 occasions this was the amygdala (3 patients: Patients 6, 7 and 12) (Fig. 2); on one occasion in the amygdala and the hippocampus (Patient 6); once in the para-hippocampal gyrus (Patient 5) and once in both the hippocampus and para-hippocampal gyrus (Patient 17; although it should be noted that in this patient there was no electrode in the amygdala). At the time of the patient's experience and description of the dreamy state, the discharge was always restricted to MTL. The amygdala was involved on only 4 occasions (in 2 patients). Patient 5 reported the dreamy state simultaneously with the spreading of the discharge from the hippocampus and para-hippocampal gyrus to the amygdala. The temporal neocortex was never involved when the dreamy state occurred.

Fig. 2

Spontaneous seizure with isolated dreamy state (Patient 12). Fast activity in the left amygdala (arrow). L, left; Am, amygdala; a MTG, anterior middle temporal gyrus, a H, anterior hippocampus; m MGT middle middle temporal gyrus; p H, posterior hippocampus; p MTG, posterior middle temporal gyrus; FG, fusiform gyrus; p ITG, inferior temporal gyrus; STG superior, temporal gyrus; p CG, posterior cingulate gyrus; IPG, inferior parietal gyrus; LG, lingual gyrus; MOG, middle occipital gyrus; Cu, cuneus; SOG, superior occipital gyrus. Each division = 1 s.

Within this group of patients, it is interesting to analyse those spontaneous seizures that occurred without producing the dreamy state:

  • For Patient 5, only one spontaneous seizure out of 6 included the sensation of déjà vécu. In this seizure, the ictal discharge initially involved the amygdala before its delayed extension into the temporal neocortex 15 s later. In 5 other seizures without dreamy state, the discharge spread occurred simultaneously into the amygdala and temporal neocortex.

  • For Patient 6, 1 recorded seizure out of 4 did not produce the dreamy state. This seizure was characterized by early loss of contact and post-ictal anomia. The ictal discharge spread immediately from the amygdala to the whole temporal neocortex.


The analysis of dreamy state comprises certain difficulties. As Gloor et al. (1982) have previously emphasized, the symptom is ephemeral and subjective: the subject lives out an experience that can be both emotional and perceptive in nature. In certain cases, the subject's capacity for remembering this experience is altered, emphasizing the importance of questioning the patient during or just after the symptom occurs.

Semiological analysis and classification of dreamy state

In this work, we have referred to Jackson's definition of ‘dreamy state’: sensations of déjà vécu, elaborate visual hallucinations and feelings of ‘strangeness’.

The mental phenomena described under the term ‘feelings of strangeness’ are rather heterogeneous. They have been reported by Bancaud et al. (1994) and Gloor et al. (1982) as a sensation of strangeness or unreality of the environment: a ‘far-away’ feeling or a feeling of someone being nearby. However, in our data and in previous reports, these did not have a mnesic character and were never associated with déjà vécu or visual hallucinations. In our view, this phenomenon should therefore be excluded from the ‘dreamy state’ definition.

In our study, the predominant sensory modality of these elaborate hallucinations was visual. For example in Case 6, the patient sees scenes from childhood and in case 12, the patient spontaneously stated, ‘I see’ in describing his experience.

The content of these hallucinations differs from one patient to another, varies between episodes of hallucination within the same patient, and even changes according to the stimulation site. Only Patient 7 consistently experienced the same hallucination: a memory of an upsetting episode from his life that took place a few hours before the recurrence of his seizures. Here we have a rare example of the possibility that the hallucination recalls an initial event associated with the first-ever seizure. It is noteworthy that most of the hallucinations consisted of people, and sometimes of the patient themselves. The patient might see himself as a child in a particular setting, thus evoking a memory, contrary to what is observed in cases of autoscopy. Sometimes characters in the hallucinations could speak, but all of the patients stated that they were unable to hear them clearly and never recalled a particular word or phrase heard. Few authors have reported cases of ‘elaborate and familiar auditory hallucination’ interpreted as human memories (Moriarity et al., 2001). This last case is in line with the classification proposed by Penfield, grouping together elaborate auditory and visual hallucinations under the term of ‘experiential hallucinations’ which he considered to be the reproduction of a past experience; in other words a memory (Penfield and Perot, 1963). Penfield even observed that elaborate visual and auditory hallucinations could co-occur; however, in most subsequent work, an association between visual and auditory hallucinations has not been found (Halgren et al., 1978; Gloor et al., 1982; Bancaud et al., 1994). Moreover, in the work of Penfield and Perot (1963), the mapping of stimulation sites eliciting visual and auditory complex hallucinations did not match: auditory responses were elicited by simulation of a limited part of the superior temporal gyrus (Penfield and Perot, 1963), in other words the auditory cortex (de Graaf et al., 2000). Visual hallucinations on the other hand were elicited by stimulation widely distributed over the temporal cortex. Moreover, the particular methodology used by Penfield and Perot (cortical stimulation performed during electrocorticography) could explain the association of elaborate visual and auditory hallucinations, by superimposing the activation of a local process (auditory response) and one generated at a distance (visual response).

This suggests, as originally conceived by Jackson and more recently emphasized by Bancaud et al. (1994), that the dreamy state should be considered as an entity distinct from other elaborate hallucinations. As shown by Penfield and Perot (1963), and reinforced by Bancaud et al. (1994), the elaborate visual hallucinations that occur in the context of the dreamy state are essentially memories. These may be childhood memories, recent memories or sometimes less well-identified reminiscences.

Penfield and Perot (1963) considered déjà vécu as an entity distinct from the ‘experiential hallucination’. While déjà vécu resulted from an altered interpretation of the present, ‘experiential hallucination’ resulted from the reproduction of a past experience. In our data, however, déjà vécu and elaborate visual hallucinations were frequently reported together in patients' descriptions.

It therefore appears that there is a semiological continuity between the phenomenon of déjà vécu and such complex visual hallucinations. Certain patients simply report that they have already experienced the scene that unfolds before them, and often spontaneously associate this with a memory. It is sometimes difficult to determine from the patients' descriptions whether a particular sensation is one of isolated déjà vécu or whether this is accompanied by a more elaborate hallucination, perhaps even for patients themselves. The sensation of déjà vécu is experienced as though it were a memory. Patient 1 spontaneously stated: ‘It's as if I am reliving an old memory’. Déjà vécu is therefore a much more powerful hallucination than a simple feeling of familiarity. Spatt (2002) argues that this phenomenon is the result of faulty interpretation, due to isolated (epileptic) activity within the recognition memory system. However, our patients did not recount a misinterpretation of the current scene; their experience of this was rather obscured by the feeling of déjà vécu, and in certain cases, the onset of visual hallucination.

Consciousness and associated signs

In our series, loss of contact was rare and occurred late. Rapid, early loss of contact is incompatible with the onset of dreamy state. This observation is somewhat different from a previous work reporting up to 28% of early loss of contact in seizure with dreamy state (Bancaud et al., 1994).

In our study, >30% of dreamy states provoked by stimulation (Table 3) were accompanied by a sensation of anguish or fear.This is in keeping with studies by Mullan and Penfield (1959) and Bancaud et al. (1994), which found such an association in 24% of spontaneous and provoked dreamy states. Fear is a frequent ictal symptom in limbic temporal lobe seizures (Cendes et al., 1994; Maillard et al., 2004). The main hypothesis to explain this association is therefore that both symptoms are produced by a partially common network, involving certain MTL structures, particularly the amygdala (Cendes et al., 1994; Biraben et al., 2001; Maillard et al., 2004). Similarly, the amygdala has been shown to process and functionally link both emotions and autobiographical memory (Maratos et al., 2001). We observed a link between the content of dreamy state and the occurrence of fear only in one case (Patient 7). This woman always had the same hallucination, consisting of a frightening episode from her childhood. In this patient, the amygdala played an essential role in the neurophysiological organization of seizures. During each seizure, the patient relives the original event including its emotional component. The impact of the initial stressful event and the role of the amygdala might explain the characteristics of this type of dreamy state (Mishkin and Appenzeller, 1987; McGough, 2002).

Anatomy of the dreamy state

The respective role of temporal lobe neocortex and mesial structures has been long debated (Mullan and Penfield, 1959; Gloor et al., 1982; Bancaud et al., 1994). In our series, dreamy states were provoked only by stimulation of MTL. All the MTL structures can produce a dreamy state when stimulated with different thresholds of sensitivity. In our study, we demonstrated that the amygdala and the hippocampus play a predominant role. Contrarily to findings of Halgren et al. (1978), an after-discharge was not necessary to elicit a dreamy state, as MTL activation could result simply from the electrical stimulation applied (17.5% of cases) without after-discharge. These results are consistent with those of Gloor's study reporting 45% of dreamy state without after-discharge. These different results are related to the different intensities and conditions of stimulation. However, the important point is that whatever the nature of the dreamy state, in our series, the after-discharge remained localized to the MTL while the clinical sequence of events took place. Spread to the temporal neocortex was rare and occurred later (at least 15 s). The comparison of these data with the relevant stimulations and the spontaneous seizures without dreamy state recorded in the same patients allow us to state that, when there is a simultaneous or early involvement of both the mesial and lateral temporal lobe, dreamy state does not occur. Bancaud et al. (1994) reported dreamy state evoked by stimulation of the neocortex, notably the superior temporal gyrus. It must be stated that in their study, the discharge often simultaneously involved the neocortex and the limbic structures. Moreover, implementation of the SEEG methodology used in their study was quite different from now, most of their recording having been performed within a single half-day recording session following general anaesthetic (for depth electrode implantation). Intensity of stimulation was also much higher than in our study. They recorded an isolated dreamy state in only 22% of cases and reported 28% of cases with early loss of contact (within the first 10 s after seizure onset). This makes it easier to understand the widespread nature of discharges reported here, contrary to our own observations and those of others (Gloor et al., 1982).

All of our patients had left hemisphere dominance for language. All of the occurrences of déjà vécu were provoked by stimulation of the right temporal lobe. Similar findings were also reported by Cole and Zangwill (1963), as well as Weinand et al. (1994), but not by Adachi et al. (1999). Gloor et al. (1982) and Bancaud et al. (1994) found no such right-sided predominance for elaborate visual hallucinations. A dreamy state was produced in 11 patients by stimulating the right temporal lobe and in 4 patients by left temporal stimulation. In one patient a dreamy state was provoked by stimulation of both the right and left temporal lobes.

Spontaneous dreamy state during temporal lobe seizures has a rather ambiguous localizing value in the literature. For Bancaud et al. (1994), the dreamy state accompanies an epileptiform discharge co-activating the limbic structures and the temporal neocortex. Our data are consistent with Gloor's data (Gloor et al., 1982) and demonstrate that the dreamy state occurs when the epileptiform discharge is limited to the MTL alone.

Phenomenology of the dreamy state

We have emphasized here the semiological continuity between ‘déjà vécu’ and visual memories. For Patient 6, the content of the hallucination is suggested by the subject of the book that she is reading. We can hypothesize that this is one mechanism that permits access to certain real and ‘vivid memory’ or reminiscence with both perceptive and emotional components, which is very different from a consciously evoked memory. A current model of autobiographical memory (Conway et al., 1999) suggests that the dreamy state facilitates direct access to such memories. The comments of patients indicate that they relive a past experience, including its personal and emotional dimensions, in other words the re-emergence of ‘autonoetic consciousness’, as defined by Tulving (1995, 2002). In our series, recall of a specific public or historical event was never reported. These hallucinations can be classed within episodic and autobiographical memory. Episodic memory is characterized by its capacity for a remembered event to be situated within a specific space and time. According to the model proposed by Conway (1996), the memory component of the dreamy state is a short-lived event that the patient can place within the context of a particular period of their life. For the patient, the memory may seem recent or very distant, sometimes from childhood. Apart from one exception, in our series, the scene involved in the ictal dreamy state varied from one seizure to another, but the memory was always from the same period of the subject's life. In certain cases, the recalled scene had probably been lived through many times during a certain period of the subject's life (e.g. Patient 6).

Penfield and Perot (1963) emphasized that experiential visual hallucinations were only observed during temporal lobe seizures, and provoked only by electrical stimulation of the temporal lobe. Our data are in keeping with those of Gloor et al. (1982), showing that memories are specifically provoked by stimulation of the MTL, and observed during those discharges limited to the mesial temporal cortex.

Lesions of the MTL can give rise to memory impairment (Squire and Alvares, 1995; Nadel and Moscovitch, 1997; Vargha-Khadem et al., 1997; Viskontas et al., 2000). The various theories of autobiographical memory and the data from functional imaging studies indicate a preferential role for the hippocampus in the recall of autobiographical memories (Vargha-Khadem et al., 1997; Mishkin et al., 1997; Tulving and Markowitsch, 1998; Nadel and Moscovitch, 2001). However, in order to construct the spatial and temporal content of the memory, the hippocampal and amygdala connections activate widely distributed cortical regions including prefrontal and temporo-frontal cortex (Buckner et al., 1999; Maguire, 2001; Maguire et al., 2001; Shastri, 2002; Greenberg et al., 2005). The dreamy state could therefore be viewed as the result of ictal activation of the MTL, the hippocampus and particularly the amygdala, which could produce the synchronization and functional activation of a wider neuronal network (e.g. in the theta band as proposed recently by Barbeau et al., 2005), giving rise to the visual, temporal and emotional representation of the hallucination. When the epileptic discharge is limited to the entorhinal cortex (Bartolomeï et al., 2004) it could give rise to a ‘pure’ experience of déjà vécu, and the visual component may be absent or unclear. Recently, Fell et al. (2006) showed the role of rhinal–hippocampal connections in the declarative memory formation during REM-sleep.

Two mechanisms of activation have been proposed to explain the dreamy state. The first revisits a Jacksonian conception, proposing a functional de-activation of the amygdala and hippocampus by the ictal discharge, resulting in liberation of neocortical structures (Halgren et al., 1978; Bancaud et al., 1994). Gloor et al. (1982) considered rather that the dreamy state was a positive phenomenon linked to the activation of MTL, or by the activation or inhibition of widespread neuronal networks (Gloor, 1990). A positive effect of activation is supported by two arguments. First, functional imaging data show an active involvement of the hippocampus in autobiographical memory (Nadel et al., 2000; Ryan et al., 2001). The second argument is the time-locked occurrence of the dreamy state with the stimulation or the epileptic discharge.

In the dreamy state, the recalled memory can be recent or remote. The mechanism involved does not appear to be different for these two types, in that the discharge involves only the MTL structures. This permanent role for the hippocampus and amygdala tends to invalidate the model of memory consolidation proposed by Squire and Alvarez (1995) but is in keeping with various other works (Giffard-Quillon et al., 2001; Maguire et al., 2001; Ryan et al., 2001).

Déjà vécu and visual memories involving recent and remote memories can be explained by the role of the amygdala, the hippocampus and rhinal cortex, whether right or left-sided, in the mechanisms of episodic autobiographical memory. The connections between these structures organize the content and the elaboration of the dreamy state for both remote childhood and recent memories.


View Abstract